June 13's award ceremony at the 8th BAAI Conference announced and presented winners for Track 1 and Track 3 of the FlagOS Open Computing Global Challenge's first season; Track 2 results are due in mid-July.
More than 1,100 developers joined the January-June 2026 competition, which BAAI and the FlagOS Community positioned as a global testbed for AI system software, chip programming and practical deployment.
Three tracks targeted distinct bottlenecks: operator development and cross-platform optimization, full-stack inference throughput for the Qwen3-4B model, and long-context automated data annotation using the same model.
Organizers said the multi-season contest carries a total RMB 2 million prize pool and is meant to expand an open-source, multi-chip computing ecosystem by drawing students, compiler developers and AI engineers into FlagOS.
Is FlagOS a path to AI independence or a clever workaround for China's reliance on US chips?
With its AI giants now closing their models, is China's state-backed open-source dream already failing?
FlagOS Open Computing Global Challenge Season 1: Over 6 Months, 3 Tracks, and a New Era for Cross-Platform AI System Software
Overview
The FlagOS Open Computing Global Challenge Season 1, held from January to June 2026, concluded at the 8th BAAI Conference and attracted developers from around the world. This event marked a major step forward for AI system software by offering thoughtfully designed, industry-oriented tracks focused on real-world technical scenarios. Participants benefited from comprehensive open-source technical support, which helped them tackle complex challenges and gain hands-on experience. The competition’s global reach and practical approach made it a pivotal moment for advancing open computing and fostering a vibrant, collaborative AI community.